In moving picture coding processing, in general, the amount of information is reduced by utilizing redundancy in the spatial direction and the temporal direction which moving pictures have. Here, in general, transform to a frequency domain is used as a method utilizing redundancy in the spatial direction. Further, inter-picture prediction (hereinafter, referred to as “inter prediction”) coding processing is used as a method utilizing redundancy in the temporal direction. In inter prediction coding processing, when a picture is coded, a coded picture that appears before or after a current picture to be coded in the display time order is used as a reference picture. A motion vector is derived by performing motion detection on the current picture relative to the reference picture. Then, redundancy in the temporal direction is eliminated by calculating a difference between image data of the current picture and predicted image data obtained by performing motion compensation based on the derived motion vector (for example, see Non Patent Literature (NPL) 1). Here, in motion detection, a difference value between a current block in a current picture to be coded and a block in a reference picture is calculated, and a block in the reference picture with which the smallest difference value is obtained is determined as a reference block. Then, a motion vector is detected using the current block and the reference block.